TLDR: Turned out that all Intel processors have flaws, and are exposed to attacks. So, there are a patch that will fix it and slow your processors. ( different websites tells different range of percentage of slowdown)
I just bought a new i9 for working with VR Game Development, Unreal Engine, Maya, 3Ds Max, Blender, Zbrush, Substance Painter, Designer, B2M, and other 3d software, VR Gaming, Rendering, Photoshop etc…
How do you think this new, slowing patch will affect my workflow? Because the range of slowdown is quite wide (5-50% slower) and I don’t know which tasks will be affected most.
Also, maybe I should return it and wait for new processors? Does anyone have i9 and can tell if there is any difference, after patch?
It’s not just Intel that has this design flaw, it’s all CPUs - Intel, ARM chips, and AMD. Intel just has the overall worse problem, but also has the widest marketshare of CPUs, so I think that’s why media outlets are addressing it as an “Intel only” problem, even though it definitely isn’t.
As for the percentage of slowdown, I can’t imagine it will affect us consumers too greatly. It will adversely affect things like servers (already seeing this with Fortnite servers), giant render farms, and probably supercomputers.
As for when it will be fixed at the proper level (the hardware level), that could take years of research. There is something wrong with the fundamental design of the chips in question, physically. Software alone can’t fix all of the problems, I imagine a lot of research and development will have to go into alternate designs for how the CPUs handle memory addresses and such.
Maybe the chips will have to become bigger again.
I think you’ll be fine with your CPU choice. Try it out and stress it out to the best of your ability and I think you’ll be pleasantly surprised. As long as you’re not planning on turning your personal rig into some kind of server/render farm, you’ll be good.
No. There are two types of attack. Meltdown (which is far easier to exploit and only apply to Intel chips) and Spectre (which is much hard to exploit).
To use Spectre attack you must run it from ring-0, and if you already have access to it… then you don’t really need to exploit it.
Patch for metldown will slowdown anything heavy on syscalls (networking, file I/O etc).
If I’m not mistaken, in the future software will need to be recompiled to avoid these security exploits, and most of the time software can flag for the CPU that certain instructions are not a security concern, so they can run normally without much impact on performance.
Anecdote: Other than the start menu on my primary Windows PC being absolutely positively broken, I notice absolutely no ill effects. i7-4790K with 32G.
For general development the performance impact is likely to be much less than it’s been on servers running in VMs and doing lots of network and disk access.
Thank you for all your answers. Yes, I know that there are a lot of panic over there, but I am spending a lot of money on my work machine, so I always want to be sure
Apparently you aren’t aware of just how many billions of dollars were spent to ensure software compliance with Y2K, starting from about 15 years prior to it. If we had had anywhere near the amount of software in play in 1985 that we do now, and as much of it was broken as it was in 1985, there absolutely would have been something more catastrophic than a few glitches here and there.
This is a very difficult to exploit flaw, and we’ll be seeing consequences from it for quite a long time. Clearly, it was very difficult to discover in the first place, and now that it’s out known, and explained, it’s still difficult to exploit. But difficulty in use has never stopped people, only slowed them down. If someone figures out a new and novel technique for exploiting it, that makes it easy, then there’s the potential of there being an absolute storm of crap to follow.
Now, as to the performance issues – it’s my understanding, that in between the time that this was discovered, and the time to patch it, no one involved has really had the time to go and perf test across many thousands or tens of thousands or even hundreds of thousands… of machine configurations, considering that the patched flaw affects literally all Intel Processors manufactured since the mid 90’s. It’s not unreasonable to consider that tripling the amount of work that the kernel has to do in certain situations could have extremely detrimental and difficult to predict effects on some hardware and software. Possibly even widespread. Also not unreasonable to consider that the faster the hardware, the less you’ll notice the effect. Well, it turns out we’re pretty lucky so far – most modern software/hardware seems to have hardly been affected by the significant increase in kernel code that is running to deal with the problem. If patches for older versions of Windows and Linux end up hitting, that run on older hardware, XP or even earlier hardware, that is still in service, then those may see significantly larger performance hits than what we see on our nice modern gaming and development rigs.
I think the worst of it is going to be mobile exploits, though. There’s no way that Google or Apple are issuing patches for every device in existence, despite Android 4-5 still being the most commonly used versions. Once someone can exploit the ARM version of the flaw, you’re going to see some serious freakouts.
